The Death of the Software Engineer by a Thousand Prompts

My wife is working on an idea for a social platform. She’s been using Lovable to build it. Over the past week, she’s made very good progress independently but once a day she would get stuck because the AI isn’t doing what she’s expecting. Unfortunately, she doesn’t know full-stack development well enough to prompt it towards resolution or how to edit the code herself. When that happens, I come to her rescue.

Verdi, the Super unblocker

Her side project gave me a sense of the future. My thesis is that AI will fragment the role of software engineering. It will become a role with a large pool of low-skilled coders who move forward with AI and a few specialists that will unblock those coders when stuck as well as address performance bottlenecks for production-scale.

Keep reading to see why I believe this and the imminent threat it creates for engineers today.

AGI is Grossly Oversold

Let’s address the elephant in the room first. None of what I will say matters if AGI being around the corner is real. The problem is that it’s not. AGI is grossly oversold. We are in peak bubble territory and no one knows when the music will stop.

What incentives do the large AI labs have? Their businesses are extremely capital-intensive. Therefore, they want you, the general public, and their investors to believe they are right around the corner of inventing AGI which will make them and their investors disgustingly rich. The stronger this belief, the longer they can keep the money flowing.

Now, these foundational AI models will continue to improve in intelligence but not in their ability to be human-like. From first principles thinking: LLMs are just fancy predictive algorithms. The fundamental weaknesses that exist today in predictive algos will still exist in the future, even if that algorithm has 100x more compute or data.

Have you noticed the goal post for AGI is constantly moving anyway? To believe “human endeavor is doomed” is to drink the Sam Altman Kool-Aid. We are far from the end of value creation by human enterprise as we know it.

Realistic Outlook: The Copilot

So what can we expect instead? A partnership with the AI systems, commonly known as Copilot. The greatest impact in the economy will be the emergence of AI tools capable of most of the generic work in most job functions doing the work while being directed by humans.

Going back to the earlier example of my wife: her AI slave bot produced over 2,000 lines of code, none of which she designed, engineered, or wrote. She simply set the vision for what to do. At critical points, a highly knowledgeable specialist (ie: me) had to intervene to get the robot unstuck and overcome a wall.

This is the future I expect to see come to fruition.

Your Salary = My Margin

Many software engineers I speak to have the wrong outlook on this matter and don’t see the danger they are really in.

“AI can never replace me”.

They say. And they are right, but the critical thing they don’t understand is:

It doesn’t need to.

It just needs to be able to deliver 80% of your output at 20% of your cost to find your role on the chopping block very fast. It makes ruthless financial sense if you think about it.

The impact of AI adoption is guaranteed to be a net displacement of labor. Where a company would previously have 30 software engineers working on a product or feature, in 5 to 10 years, they will have 20 AI prompters and only 10 software engineers.

History repeats itself. In the last century, typesetting used to be a big industry. It once required specialized skills and machinery. You could make great money being one. However, in the 1980s, desktop publishing software suddenly enabled anyone with a computer to design and print content. This democratized publishing led to a decline in traditional print jobs and a rise in graphic design and DIY publishing in its stead.

Software engineering (and most knowledge work) is facing a similar fate.

The Path Forward

The upside is that software engineers of today will transition to roles that are higher-level with more visibility, seniority, and scope. Akin to a doctor working with many nurses under them. The downside is the number of such roles will be dramatically slashed in comparison.

The way to survive in this changing world is to stay ahead of the curve. Those who thrive in the transitions will have one of two things going for them, or ideally both:

• They become seriously great at architecture and system design to gatekeep the subpar code the AI coders will pump out.
• They invest heavily in soft skills since the ability to lead and manage groups of semi-coders will become more valuable.

All in, the future looks bright for the opportunist but bleak for the complacent.